When you practice a skill in your head it’s clearly different from practicing it physically— you’re not actually doing it! Yet research has shown time and time again that you can learn a skill by practicing mentally, even if you’ve never done it before.
Simulating a movement mentally is known as motor imagery. It’s a popular topic in movement neuroscience for a number of reasons. For one thing, it has a lot of potential applications in sports training and rehabilitation. And on a fundamental level, it’s just plain interesting—how is it that we can learn a movement skill without actually practicing it physically? This question was one of the motivations for a paper I recently published in the journal Behavioral Neuroscience.
Researchers are still unsure as to whether skill acquisition via motor imagery is due to improvements in perceptual processing (like learning to pick out relevant visual information), movement planning, or execution of the movement. Not surprisingly, it seems that learning relies on a combination of many different processes. But does motor imagery rely on certain processes more than others? And how does motor imagery compare with physical practice?
For this study we aimed to provide some insight using transfer tasks. Using a transfer task simply means that you test participants using a task that differs from training somehow. Doing so allows you to test the ability of the training to “transfer” to a new situation. For example, you could have people train at a particular speed, then test them at a new speed and see how well they do.
In our study the training task was to practice a sequence of button presses either through motor imagery or physical practice. Participants sat at a keyboard and heard random numbers from 1 to 4, and they responded by pressing the appropriate key— one group actually pressed buttons, the other imagined pressing them. To the participant, it sounded like a bunch of random numbers, but in fact we had embedded a repeating sequence. If people typed out the repeated sequence faster and more accurately compared to random numbers, then they learned the sequence!
After training, all participants performed one of two transfer tasks: a “perceptual transfer” and a “motor transfer”. In the perceptual transfer condition, we changed the audio cue to a visual one; the numbers appeared on the screen in front of them instead of hearing them. And for the motor transfer condition, we had participants switch hands; they practiced with their left hand, and were tested on their right.
What we found was quite interesting: motor imagery based learning was disrupted by changing the cue, but not so much with changing the hand. Physical practice, on the other hand (pun intended), was affected by both changing the cue and changing hands — but changing hands had a greater effect.
What might this mean?
We concluded that learning via motor imagery might rely more on perceptual processes (like learning how map the correct response to a particular cue) compared to physical practice, which appears to rely more on motor processes (actually improving execution of the movement). This lines up well with research showing that motor imagery consistently activates brain areas involved in perceptual processing (e.g. the parietal cortex) but less reliably activates brain areas involved in motor processing (e.g. the motor cortex).
That doesn’t mean motor processing isn’t involved in motor imagery — it could be, especially with practicing tasks that rely more on motor processes by their complex nature. Sequence learning tasks use button presses… and if you use a computer regularly, you can’t get much better at executing button presses in a short experiment like this. Therefore these types of experimental tasks likely rely more on perceptual processes and motor planning, like mapping a response (the movement) to a stimulus (the cue) — which is still an important aspect of motor learning. Imagine a baseball player who could swing a bat perfectly as planned, but keeps picking the wrong swing for a given pitch!
Either way, our results show that (at least with this type of task), motor imagery based learning seems to depend heavily on perceptual learning or planning processes compared to practicing physically. This might be one of the more important differences between motor imagery and physical practice.
As with most studies, this leaves us with more questions than we started. Personally, I’m very interested in whether this is true in later stages of learning (more on that in another post), and whether this applies to more complex tasks that rely on different processes. And since motor imagery appears to drive learning in a way that is fundamentally different, could combining it with physical practice have unique benefits? This is something I’ll blog about later… but for now, I’m done pressing buttons!
 Rienzo, F., Debarnot, U., Daligault, S., Saruco, E., Delpuech, C., Doyon, J., … Guillot, A. (2016). Online and Offline Performance Gains Following Motor Imagery Practice: A Comprehensive Review of Behavioral and Neuroimaging Studies. Frontiers in Human Neuroscience, 10. doi:10.3389/fnhum.2016.00315
 Kraeutner, S. N., MacKenzie, L. A., Westwood, D. A., & Boe, S. G. (2015). Characterizing Skill Acquisition Through Motor Imagery With No Prior Physical Practice. Journal of experimental psychology. Human perception and performance, 42(2), 257. doi:10.1037/xhp0000148
 Ingram, T. G., Kraeutner, S. N., Solomon, J. P., Westwood, D. A., & Boe, S. G. (2016). Skill Acquisition via Motor Imagery Relies on Both Motor and Perceptual Learning. Behavioral neuroscience. doi:10.1037/bne0000126
 Some examples: improving perceptual processing could mean one improves their ability to pick out relevant features from a cue — features relevant to how you’re going to respond with movement — or discriminating between similar sensations. For example, learning what aspects of a pitchers movement helps predict where the ball is going. Improving movement planning might mean learning how to select the best response to a given cue — perhaps the best baseball swing for a given pitch. Improving movement execution might mean performing the actual movement faster and/or more accurately — getting better at swinging the bat exactly how you want. As you can imagine, these processes interact and are very hard to tease apart experimentally.
 And we only included people in our final analysis if they didn’t realize there was a sequence until we told them at the end — that is, they learned “implicitly”, without explicit awareness of the sequence. This way we could avoid (somewhat) having people simply memorize the sequence of numbers like remembering a phone number, and tuning out during the random numbers. We wanted people to perform both situations as similarly as possible.
 Note that all participants, whether they practiced physically or using imagery, performed this final transfer test physically… or else there would be no data to record in the imagery group!
 By the way, we already published a study demonstrating what happens when you keep the test conditions exactly the same as training (not a transfer task), so we already had baseline data. If you want to check that out, see Kraeutner et al. (2015).
 Hétu, S., Grégoire, M., Saimpont, A., Coll, M.-P. P., Eugène, F., Michon, P.-E. E., & Jackson, P. L. (2013). The neural network of motor imagery: an ALE meta-analysis. Neuroscience and biobehavioral reviews, 37(5), 930–49. doi:10.1016/j.neubiorev.2013.03.017
 Yes, you could get better at typing quickly if you put your mind to it. But keep in mind that would involve learning to type entire words and sentences faster, learning to transition between keys better, mapping the keyboard more accurately, etc… but in this study, you’re just hitting one of four keys. Right now, you’re probably as good at that as you will ever be.